An iterative design of experiments based data collection approach for ultrasonic guided waves

To validate a model of guided wave propagation from acoustic emission on real aerospace structures a large amount of experimental data must be collected. It is desirable to take measurements over large areas at a small spacing between measurement points and this is very time consuming. To reduce the experimental eort, the HilomotDoE algorithm has been adapted to provide a black box method of selecting the data points to be collected. It chooses points which do not fit well to local hinging plane models. For example this includes the drop in amplitude of a guided wave as it crosses a stringer. The algorithm has been tested on a simulated structure and experimentally on a real structure. On the simulated structure, which had a relatively low feature density, using the algorithm reduces the number of data points required to 17.3% of the number of points in a full raster scan to reach a suciently good representation of the full raster scan. On the experimental structure, the algorithm did not reduce the number of points collected significantly and this was due to the much higher feature density in this structure. This highlights that the algorithm will only reduce the number of collection points required where appropriate. This algorithm has potential in reducing the number of data points collected in other
NDT applications where measurement time is large and parts of the response are predictable.